PM2E Hang (fwd)

MegaZone (megazone@livingston.com)
Wed, 11 Dec 1996 17:41:26 -0800 (PST)

Once upon a time Kevin Fink shaped the electrons to say...
>One of our PM2Es (2 modem modules, one 5-BRI-U module, running ComOS
>3.3.3, SNMP disabled, 4 MB memory) crashed today. It would not respond to
>pings, telnet, or any ports (including S0, which I always leave set to the
>diagnostic mode). Unfortunately, I didn't have a terminal on S0 at the
>time, so didn't get any information when it crashed.
>
>I did notice that the ISDN lights were all behaving normally (i.e. the one
>which is actually connected to an ISDN line was steady on and the rest
>were flashing quickly, which isn't what the manual says should happen, but
>is anyway...)
>
>Also, the ethernet lights were behaving normally (green on, red flashing
>almost continually).

You know, I was about to say that this sounds just like what happens with
spurious interrupts...

>The only thing I received via syslog were a series of "spurious
>interrupts" errors just before it stopped responding.

That was the problem. The spurious interrupts caused the CPU to race
and not have time to service the ports.

>1) Is it considered bad practice to unplug or plug in serial cables
>(attached to turned-off modems) while the PM is on? We have always done
>this, but if it is going to crash the terminal server... I'd rather not
>shut down the whole server just to swap out a modem, though.

That shouldn't cause a problem, I don't think that would be it. Maybe
he managed to cross pins to cause the interrupts? Or the swap caused a
modem to fritz and generate them?

>interrupts. I am not running SNMP. Could that have allowed me to collect
>more information or is it only useful for normal operation?

Doubtful. Spurious Interrupts are like 'while (1); fork', the unit is
preoccupied trying to service them to the point it grinds to a halt.

<!-- Distribution:World Status:Verified Author:megazone -->
Releases prior to 3.3:

Spurious interrupt means that CTS and/or DCD startled toggling (low to
high over and over) more than 10,000 times in a one second window.
Either one or both of those signals toggling like this will cause a
Spurious Interrupt.

Release 3.3 and up:

Serial port spurious interrupt handling has been extended to include
detecting streams of framing errors. Some modems get confused about
their configuration and begin sending continuous data to the
PortMaster at a baud rate different than set on the PortMaster. This
would cause all operation on the PortMaster to appear stopped for
several minutes to several hours. The PortMaster now attempts to
reset the modem and will continue to operate properly even if the
modem does not recover.

-MZ

--
Livingston Enterprises - Chair, Department of Interstitial Affairs
Phone: 800-458-9966 510-426-0770 FAX: 510-426-8951 megazone@livingston.com
For support requests: support@livingston.com  <http://www.livingston.com/> 
Snail mail: 4464 Willow Road, Pleasanton, CA 94588